First, isn't there a better word for consciousness, as used in this book? Because I have the feeling that different people (even scientists) mean different things by it.
Second, I think the main problem is that most people consider consciousness as something which lies outside of the physical realm, and that there is, supposedly, a one-way communication link from the physical world to our consciousness. However, since we are in fact (physically) talking about the concept of consciousness, this seems highly unlikely. It is thus more likely that consciousness is a part of physics, and we should perhaps treat it as such.
I don't remember the book (or this review of the book) ever suggesting consciousness existed outside of the realm of physics or the natural world. In fact, the opposite: it tried to show how consciousness might have come about through natural means (feedback loops).
This is the standard reductionist position, and the standard strawmannirg of people who don't agree with it.
To explain it crudely, yes, it all ultimately reduces to the neuronal substrate and no one in their right mind disagrees. The point is that only looking at this level ignores the important and large set of data that is humans' first-hand observations of consciousness, what you're able to perceive, study, and observe about conscious experience, the kind of observations that Searle makes. To just write off first hand observation data as irrelevant strikes me as anti-scientific if not ideological. I mean, why wouldn't you try to develop a system of consciousness that starts with this data, in tandem with neuronal studies (there's no inherent contradiction)?
The problem is that first-hand observations aren't irrelevant, they're perverse. They're the mind viewing itself through its own intuitive folk-theory of mind, which is nowhere near an accurate psychology.
> They're the mind viewing itself through its own intuitive folk-theory of mind
Why do you think I'm suggesting any theory of mind at all? You're stating too much.
I'm just suggesting look for patterns and produce models of what you can observe about human subjectivity. Hofstadtr's "strange loop" being an example of one such model.
>I'm just suggesting look for patterns and produce models of what you can observe about human subjectivity.
Well yeah, but the whole problem with subjectivity is that it consists in the mind's model of the environment feeding back into the sensory inferences as inductive bias while the feature data extracted from the sensory inferences also trains the model of the environment.
Which is why people's first-hand "observations" aren't really observations in the scientific sense: "sensings" or "appearances" are actually theory-laden inferences.
That's a narrow viewpoint. Sociology, history, psychology, psychoanalysis, philosophy, even the hated field of literary theory all provide fertile ground for drawing conclusions about the nature of human subjectivity. The people who claim, apriori, "you can't model human subjectivity and we shouldn't bother" are not going to be on the side of progress, especially with regards to AI.
Oh don't get me wrong, I'm all for AI and I think we can do it. I just think it is inherently impossible to measure subjectivity. Probably. Unless we do discover some mechanism for it, but all the things you describe (psychology, philosophy, etc), none of them give any evidence for subjectivity. Only that things act like they are conscious.
I'm just on the side that if something looks and acts like it is conscious, we have to assume it is. But the only being anyone knows is conscious is themselves.
I have a short article that covers some similar topics. In the third paragraph I make a strong argument against materialism by examining consciousness.
> it seems disingenuous to either believe that many separate things can have a synthesized experience.
This isn't an argument, it's merely an assertion of a point of view, and one that isn't obviously true. The evidence on the other side of the argument seems much stronger than your argument about our visual field. When you look at all of the examples of brain damage causing peoples personalities to change dramatically, well you have a high hill to climb to say that who we are isn't material in nature.
the quoted text foreshadows the argument about visual fields. I'm not saying that the brain isn't important, just that it cannot solely explain our conscious experience. This alone is sufficient to defeat materialism.
Ghassan, I was intrigued by your article. I wanted to share with you some thoughts I had regarding it, via email rather than here. But the email address you post on your website is broken. Your domain lacks an MX record. You might want to fix that.
"We share a conviction that no philosopher or
scientist living today has the foggiest notion of
how consciousness, and its inseparable companion
free will, emerge, as they surely do, from a material
brain."
That's reasonable enough. Nobody seems to have a clue about how "consciousness", whatever it is, really works. If someone did, it would have been implemented in a computer by now.
"We mysterians are persuaded that no computer of the sort we know how to build—that is, one made with wires and switches—will ever cross a threshold to become aware of what it is doing."
Nah. Now he's just mouthing off. There's a long history of this sort of thing from philosophers. Aristotle claimed humans were special because they could do arithmetic. Now we know that arithmetic is simple, in the sense that it takes a very small parts count. Hubert Dreyfus, in 1972, said computers couldn't play even a decent amateur game of chess. Now, any laptop PC can trounce grand masters.
What AI isn't doing well yet is predicting the consequences of actions: "what will happen next", and "what will happen next if this decision is made". Basically, the technology to get through the next 15 seconds of life. This is the beginning of common sense, and really ought to be worked on more. We have the tools now to go at that problem. Self-driving cars have to do some of this, and answers may come out of that technology. This is an essential technology for robots which operate in an unstructured world.
Once that problem has been solved, we might be in a position to talk about "consciousness" from a non-clueless perspective. Higher level thinking is mostly a strategy module for the "next 15 seconds" control system.
At least the article doesn't use the world "quantum".
Higher level thinking is model building from sensory input. I have not seen any system out there that learns to reason about 3D space by just going around touching/feeling things and then incorporating that feedback with other senses. Everything starts with a model built-in and then tweaks parameters through some kind of optimization mechanism. Which is fine but that's not really intelligence in the sense of coming up with novel models and incorporating it into pre-existing models.
I'm expressing the difference from going from 2D Euclidean geometry to 3D Euclidean geometry to Non-Euclidean geometries as in the development of all of those models and theories throughout the history of mathematics. That's what I mean by model building. There currently is no such artificial system that can go through the process of developing and evolving such models and systems. So reasoning/thinking/consciousness or whatever people want to call it must account for such model building. All current research is basically just tweaking something until it fits good enough in a very limited domain. There is no explanatory power in any of the current research.
Absolutely not. You can, in your house, build a robot that uses sensory input to construct a map of its surroundings and then navigate that map.
Check out various algorithms to do so, such as [1]. Model building from sensory input is very simple, and in my opinion only tangentially related to consciousness.
Former neuroscientist here, whose lab studied consciousness. I'm not sure about the book, but the review is light on details.
We know biology is connected somehow to conscious awareness (sleep, illusions, blindsight, TMS, etc.) but we still haven't jumped the gap separating the objective from the subjective. It's called the Hard Problem for good reason.
I tend to think subjective consciousness is just one of those fundamental forces/components of the universe, like space-time. It's so different than anything else we observe, I don't know how we ever bridge that gap between the material and the subjective. It just is.
My impression of I Am a Strange Loop was that it contains a subset of the ideas in GEB, and lacks the off the wall charm (and delicious mathematics) that GEB contains. I sometimes recommend it to people who tried to read GEB and gave up, but I don't think there's much to learn from I Am a Strange Loop if you're already familiar with Hofstadter and GEB.
I've heard that it's a much more refined version of the strange loop idea presented in GEB, although now that I think of it, it was probably the author who claimed that.
51 comments
[ 3.6 ms ] story [ 131 ms ] threadBTW, I really liked "The Mind's I" (coauthored with Daniel Dennet).
Second, I think the main problem is that most people consider consciousness as something which lies outside of the physical realm, and that there is, supposedly, a one-way communication link from the physical world to our consciousness. However, since we are in fact (physically) talking about the concept of consciousness, this seems highly unlikely. It is thus more likely that consciousness is a part of physics, and we should perhaps treat it as such.
To explain it crudely, yes, it all ultimately reduces to the neuronal substrate and no one in their right mind disagrees. The point is that only looking at this level ignores the important and large set of data that is humans' first-hand observations of consciousness, what you're able to perceive, study, and observe about conscious experience, the kind of observations that Searle makes. To just write off first hand observation data as irrelevant strikes me as anti-scientific if not ideological. I mean, why wouldn't you try to develop a system of consciousness that starts with this data, in tandem with neuronal studies (there's no inherent contradiction)?
Why do you think I'm suggesting any theory of mind at all? You're stating too much.
I'm just suggesting look for patterns and produce models of what you can observe about human subjectivity. Hofstadtr's "strange loop" being an example of one such model.
Well yeah, but the whole problem with subjectivity is that it consists in the mind's model of the environment feeding back into the sensory inferences as inductive bias while the feature data extracted from the sensory inferences also trains the model of the environment.
Which is why people's first-hand "observations" aren't really observations in the scientific sense: "sensings" or "appearances" are actually theory-laden inferences.
Actually, there exists only a data set of one. Since I'm the only one I know is conscious. Joking, but not joking.
I'm just on the side that if something looks and acts like it is conscious, we have to assume it is. But the only being anyone knows is conscious is themselves.
http://www.gmw6.com/belief_through_reason.html
This isn't an argument, it's merely an assertion of a point of view, and one that isn't obviously true. The evidence on the other side of the argument seems much stronger than your argument about our visual field. When you look at all of the examples of brain damage causing peoples personalities to change dramatically, well you have a high hill to climb to say that who we are isn't material in nature.
You obviously haven't taken much drugs! Some do in fact make you smarter.
"We share a conviction that no philosopher or scientist living today has the foggiest notion of how consciousness, and its inseparable companion free will, emerge, as they surely do, from a material brain."
That's reasonable enough. Nobody seems to have a clue about how "consciousness", whatever it is, really works. If someone did, it would have been implemented in a computer by now.
"We mysterians are persuaded that no computer of the sort we know how to build—that is, one made with wires and switches—will ever cross a threshold to become aware of what it is doing."
Nah. Now he's just mouthing off. There's a long history of this sort of thing from philosophers. Aristotle claimed humans were special because they could do arithmetic. Now we know that arithmetic is simple, in the sense that it takes a very small parts count. Hubert Dreyfus, in 1972, said computers couldn't play even a decent amateur game of chess. Now, any laptop PC can trounce grand masters.
What AI isn't doing well yet is predicting the consequences of actions: "what will happen next", and "what will happen next if this decision is made". Basically, the technology to get through the next 15 seconds of life. This is the beginning of common sense, and really ought to be worked on more. We have the tools now to go at that problem. Self-driving cars have to do some of this, and answers may come out of that technology. This is an essential technology for robots which operate in an unstructured world.
Once that problem has been solved, we might be in a position to talk about "consciousness" from a non-clueless perspective. Higher level thinking is mostly a strategy module for the "next 15 seconds" control system.
At least the article doesn't use the world "quantum".
Check out various algorithms to do so, such as [1]. Model building from sensory input is very simple, and in my opinion only tangentially related to consciousness.
[1]https://en.wikipedia.org/wiki/D*
This is one of the basic activities AI researchers have been solving for years.
In many cases, this is not even a hard problem.
For humans, the model is bound externally in parental memory, media influences, and impersonal record systems.
But it's very much there as external programming. Baby humans don't very well when it's missing.
OK.
> Higher level thinking is mostly a strategy module for the "next 15 seconds" control system.
I'm not convinced we have even that much of a clue about consciousness.
>Penrose, for example, thinks the mystery may yield to a deeper understanding of quantum mechanics.
isn't this guy's bias. he's, in part, interested in how the AI works, because he's interested in understanding how people think.
i'm also totally sick of the buzzword.
to reveal my bias, here's another thing AI doesn't do but we do all the time: have feelings.
We know biology is connected somehow to conscious awareness (sleep, illusions, blindsight, TMS, etc.) but we still haven't jumped the gap separating the objective from the subjective. It's called the Hard Problem for good reason.
Common Lisp’s loop is so powerful that people invented functional programming so that they’d never have to use it.
G_Morgan in reddit
http://www.reddit.com/r/programming/comments/a481l/so_to_get...
https://common-lisp.net/project/iterate/
https://en.wikipedia.org/wiki/G%C3%B6del,_Escher,_Bach
Legends ever say that the first man who tried to make primates to understand the square root of 2 was drowned by them.
The playful discourse evident in GEB was no more.